Privacy Preservation of Business Forecasting Using Homomorphic Encryption

Data privacy is very much essential in this digital world. Data privacy prevents the information of an organization from fraudulent activities such as hacking, phishing, and identity theft. Machine learning is an emerging technology. But a huge amount of data is required for training the Machine learning model. When an organization wants to analyze their profit rate it has to send its data to third party which may reveal organization's business tactics or sensitive data. Hence, there is always a risk of data privacy. So, privacy preserving is used. Privacy preserving prevents data leakage from machine learning algorithms. There are many privacy preserving machine learning strategies which are used for data privacy. Homomorphic Encryption is one such technique. In homomorphic encryption, the data to be fed to train the machine learning model is encrypted. The encrypted data is then fed to the machine learning model. The machine learning model performs the required computation and returns the result in encrypted form, which on decryption returns the required output

I. Introduction

Machine learning is at the heart of Artificial Intelligence. It gives machines the ability to learn things by themselves by training them. Datais very critical in Machine Learning. Privacy of the data is also very important. If an organisation wants to give its data to a third party so that it would make a prediction, then the data of the organisation might be at risk. The organisation might not want to give its data to the third party. Similarly, the third party might not want to share its Machine Learning Algorithm for maintaining competitive advantages. Even when the data or features extracted are sent through secure channels, there is a chance for reconstruction attacks, membership inference attacks [1]. Privacy Preserving is thus used to protect the confidentiality of the data by predicting the output of